Farming with Artificial Intelligence

HOST:

You’re hearing and reading about it everywhere these days: the potential impacts of artificial intelligence, or A.I., on industries across life as we know it, from packing and shipping to steel fabrication, self-driving cars, virtual assistants, and even a Fit Bit-like device for a cow. 

You may be wondering why a cow needs a pedometer, but researchers have utilized A.I. in agriculture for years to help monitor crops, soil, livestock and weather damage. They collect and analyze the data to help farmers better understand what’s happening on their farms and reduce risk to crops and livestock.

Today on Farms, Food and You, we’ll dig deep into A.I. and other technology utilized in agricultural research with Jevon Smith, a research computing manager for NC State College of Agriculture and Life Sciences. We’ll learn how data is being gathered across NC State research stations and then analyzed and processed back at NC State to help improve agriculture and predict trouble at its first onset.

HOST:

NC State CALS covers a lot of ground in agriculture research. One of the biggest challenges is connecting all that research to stakeholders in the college and across the state. Smith and his team play a major role in that.

SMITH:

I support any of the agriculture and life sciences systems, pipelines, digital assets technologies, all of our research facilities on campus, like the Plant Sciences Initiative, the Phyto Tron, for instance. We also support the 18 research stations across the state of North Carolina as well. Externally in research entities like the Plants for Human Health Institute and the NC Food Innovation Lab as well.

So it’s a lot of area, a lot of geographic space to cover, a lot of moving pieces. We have a lot of interactions also with our Extension IT partners, too. So basically seeing that pipeline go from research to teaching to Extension. We’re injected in every part of that process from beginning to end. That covers basically computing systems, what we call endpoint nodes, right? So that’s research machines that are connected to, it could be from instrumentation, like gel docs, sequencers, any kind of analytics machines, gps. So anything that’s in the lab we support. That also extends to systems as well. So, for instance, in the Plant Sciences Building, we have a high performance computing cluster that basically does a lot of our artificial intelligence and machine learning and system analysis.

So if you kind of look at the Plant Science Initiative, the Plant Sciences Building, kind of being the inside of the hub, and then the research stations are like the spokes. And so a lot of the data that we’re getting from the field comes into the Plant Sciences Building for analysis. And so we support the hub and the spokes out in the field as well. 

HOST:

One of the biggest initiatives Smith and his team are tackling in the next few years is connectivity: getting everyone internet access to better share and exchange information in real-time.

SMITH:

Really phase one is connectivity. So, obviously the federal government is doing a lot of work with this, with the new B grant and the FCC and the NTIA, which is basically two of the major components to this, the money and funding that’s coming to the states. If you talk to most farmers, even NC State farms, our biggest challenge is connectivity. A lot of farms are in very rural areas, which doesn’t help. You know, a lot of internet service providers like Spectrum and AT&T, they’re looking at return on investment.

So if they’re gonna run fiber, it costs a lot of money to deploy, as you probably know. So when they deploy it, they don’t normally look at farms and rural areas because there’s not a lot of density of people and populations there. So the return on investment sometimes doesn’t add up. So we’re really having to partner with the providers like Spectrum, AT&T and MCNC is our major partner. We are partnering with them at all levels to make us part of the process for inclusivity when it comes to connectivity for broadband. So, we are on a really aggressive plan to get all of our research stations connected to fiber over the next two to three years. Farmers obviously are gonna have a harder time because we obviously have, you know, relationships with some of the ISPs, but we’re also lobbying for connectivity to any farmer through Extension partnerships. We have a resource available, and there’s resources available at the Friday Institute and at the state level for any farmer who’s looking into connect their farm to high speed internet services.

HOST:

The next phase for his team is determining what to do with the information from research programs around the state.

SMITH:

So connectivity is number one, right? So once the connection’s made, then what, right? So then it’s how do you create pipelines? How do you create analytics pipelines? Connecting the farms and getting the data is just one piece. Once you have all this data, what do you do with it? It doesn’t have a lot of value if it’s just sitting in a database or in some kind of table somewhere. If it data is sitting in a spreadsheet or sitting in a SQL library, it doesn’t really do much. It’s how do you analyze it, how do you manipulate it, how do you get inside, how do you draw conclusions from it, right?

That’s closing the loop, what we consider it to be closing the loop. So once the data gets into the system, we have tools like SAS, that’s where SAS comes in. SPSS R, the Data Sciences Academy. So for instance, in our PSI building, we have three, I’m sorry, two dedicated think tank labs that are basically designed to now process and analyze the data, to draw conclusions and to make insights. So that’s kind of the next phase after the connectivity happens, is now helping farmers with the proper analysis of that data.

HOST:

So how important is A.I. and its analysis for farming? It enables real time mitigation for issues in the field.

SMITH:

So AI is just another level, I would say another framework that’s gonna be really important over the next 10 to 20 years. You already see it popping up in other areas like transportation, speech recognition, analysis, and agriculture. It’s gonna help make even more intelligent decisions quicker, too. So more real-time decisions. And so one of the main drivers right now is phenotyping, so identification of weeds, for instance, right? So AI can help improve the return time to help treat a field that has a weed population in it that you want to treat and remove.

So instead of waiting for the weeds to grow and, you know, take up over field, now as your tractor moves through the field, our AI cameras can actually scan and identify quickly within seconds, different weed variations and types in the field without having to have a person go out and actually manually physically inspect those. And so that’s an example of where now you’ve really changed the game for a farmer who is struggling with crop yields. So tchnology is directly helping prove that that return on investment, that they’re, they’re putting in farmers. Farming is very hard, right? Traditionally. So this tech is really helping the farmers become more efficient and productive.

HOST:

A.I. is happening close to campus at the Lake Wheeler Road Field Laboratories. Only 5 miles from downtown Raleigh, the 1,784 acres of farmland house 23 educational units that work to make farming practices more efficient, sustainable and profitable for today and generations to come. 

SMITH:

Lake Wheeler is a great asset for the university, for one, because it has really close physical proximity to the NC State campus. So students don’t have to drive, you know, 50, 60 miles away to do research, so they can go to Lake Wheeler. But, so that’s one of the geographical assets that we have. It’s a great test bed for anything that is not off the shelf and ready to go. 

So we have a lot of instrumentation, for instance, in the dairy area where we’re doing animal behavior health analysis for the cows. So the same cows that provide milk for Howling Cow, we’re actually doing research on those cows as well. It’s basically like a Fitbit device or an Apple Watch device around the cow.

It has a pedometer, so it could tell how many steps the cow has taking. It could tell when they’re ready to be milked. It has a bunch of insights in it that can help do predictive analysis. And in the future, it can even help predict and prevent disease earlier on. So the goal is to try to get in front of some of those types of problems with production of milk and other products from dairy. So, and then that’s the animal side. So in the field and food side and plant side, we have sensors in the field that do everything from soil moisture to salinity to pH, all types of different variable sensors that can pick up environmental factors, optical sensors for light photoable take. And so instead of having a person to drive out there and actually manually take these samples, it’s automatically being done by technology.

It’s all types of variable sensors that are out there, and it’s growing every year. The market’s really starting to come alive. And then the cool part is that stuff that doesn’t exist that we can’t get off the shelf, we make, so we have the Makerspace here, which is a huge asset. We actually have our grand opening on April 18th. So we now have the capability inside the college to design all the our own technology, so we can actually deploy that. So it’s even almost has a little bit of a inventorship aspect to it, too.

HOST:

NC State researchers are using A.I. to tackle all sorts of challenges.

SMITH:

Some of the largest users of this are Dr. Jean Ristaino, with her plant sensor technology, so developing sensors to prevent plant pandemics. So preventing the next famine. Dr. Chris Reberg-Horton, is using that lab for developing high-output phenotyping systems for plants and crops.

HOST:

So what’s next in farm tech? Smith has a preview.

SMITH:

So I think what’s gonna happen really in the next five years, we’re gonna reduce the time from problem acquisition to corrective action remediation is gonna really reduce.

So we’ll be able to tell you, “Okay, right now we got it. Right now, we need to spray that field immediately because there’s a problem and we can detect it at the micro level, cellular level.” We have some researchers doing stuff that are at the DNA level, RNA level, and so we’ll be able to really detect problems much earlier. And that’s just gonna be an enabler to improve in food production, making crop yields a lot better than what they are now. Obviously there’s a lot of land that’s turning more urban, less rural, so we’re, we have less land, more people. And so I think these tech technologies are gonna really help improve a lot of that cycle of problems that we can be going. 

And so I think the next big thing that we’re really focused on is the food and animal systems. We talk a lot about plants, but there’s also too a lot of tech that you really haven’t heard of that’s gonna be really important for the Food Animal Initiative that’s coming out. So I say lookout for the Food Animal Initiative, that’s gonna be huge. So we already have some researchers starting to kind of scratch the surface with animal, animal behavior health studies. And so the tech on those is gonna be really, really important. 

HOST:

To learn more about A.I. and research at work, check out the CALS Spring Magazine online.

Thank you for joining us on Farms, Food and You. This podcast is a product of NC State Extension and the College of Agriculture and Life Sciences at North Carolina State University. If you would like to support the show, please share this episode on social media and leave a review on your podcasting app of choice. Let’s talk soon!